采用改进人工蜂群算法实现基于MP的信号稀疏分解  

Signal MP-Based Sparse Decomposition with Modified Artificial Bee Colony Algorithm

作　　者：刘继承 郭睿[1] 祁春雷[1] 

机构地区：[1]东北石油大学电气信息工程学院,黑龙江大庆163318

出　　处：《自动化技术与应用》2016年第5期54-58,共5页Techniques of Automation and Applications

基　　金：教育部新世纪优秀人才计划(编号NECT-09-127)

摘　　要：针对匹配追踪信号稀疏分解的巨大计算量问题,在具有全局搜索能力的人工蜂群算法基础上,采用了一个改进的人工蜂群算法实现信号的稀疏分解。利用全局最优解和个体极值改善人工蜂群算法中的搜索方式,并引入异步变化学习因子,进一步平衡全局搜索和局部搜索的比重,改进后的算法可以快速搜寻匹配追踪过程中每一步分解的最佳原子,有效提高信号稀疏分解的速度。实验结果证明,基于改进人工蜂群的MP算法对信号稀疏分解的速度优于传统MP算法,并获得良好的重构效果。Aiming at the problem of a great amount of calculation in matching pursuit signal sparse decomposition, a modified artificial bee colony algorithm is proposed based on the global search ability of artificial colony algorithm for matching pursuit decomposition of signal. Using the global best solution and previous best solution to improve artificial bee colony algorithm in the search mode, and the introduction of the asynchronization variable learning factor further balances the proportion of the global search and the local search. The modified algorithm can search for the best atom of each step in the process of matching pursuit quickly, effectively improving the speed of signal sparse decomposition. The experimental results show that the speed of signal sparse decomposition based on the MP algorithm with modified artificial bee colony is better than the conventional MP algorithm, and obtaining good reconstruction results.

关 键 词：稀疏分解 匹配追踪 人工蜂群算法 异步变化学习因子 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]